Summary
- Dublin-headquartered Disseqt AI has raised $8m in a seed round led by Gresham House Ventures, with participation from Sure Valley Ventures, Delta Partners and Enterprise Ireland.
- Its platform combines testing before deployment, monitoring during operation and evidence intended to support governance assessments.
- The company plans international expansion, including its UK business, but independent evidence of performance across its customer base remains limited.
Dublin-headquartered technology company Disseqt AI has raised $8 million to expand software that tests and monitors enterprise AI agents, as businesses seek stronger evidence that automated systems behave within their intended operational and regulatory limits.
The seed round was led by Gresham House Ventures, with participation from existing investor Sure Valley Ventures, Delta Partners, Enterprise Ireland and individual backers. Disseqt intends to use the capital to develop its assurance technology, expand enterprise customer relationships and strengthen operations across its international markets.
Founded by Apoorva Kumar, Manish Atri and Cyril Treacy, the company operates from Dublin, London and Bengaluru. Its platform is designed to connect AI applications with the teams responsible for testing, security, risk management and compliance, allowing organisations to examine how agents behave before and after deployment.
The investment was confirmed by Sure Valley Ventures on 17 November and reported more widely the following day. The investor says the company’s sales have increased by 200% since its original pre-seed investment approximately a year ago, although it has not provided the underlying revenue figures.
AI agents require controls beyond model testing
Enterprise AI agents can perform activities involving several connected applications, including retrieving records, interpreting requests and initiating actions within business systems. Their usefulness depends partly on the ability to respond to changing information, rather than following every step of a fixed software procedure.
That flexibility also makes assurance more complicated. Testing an AI model’s answers under selected conditions does not necessarily establish how an agent will behave when connected to live data, external tools or applications capable of modifying important records.
An agent assisting with customer administration, for example, may require access to contact records, account information and communication systems. Its permissions should support the intended work without allowing unauthorised disclosure or changes to information outside its assigned responsibilities.
Failures can arise from more than the model itself. Incorrect access permissions, unreliable business data, unexpected instructions and errors in connected applications may all affect an agent’s behaviour, requiring assessment across the complete operating environment.
Disseqt’s software is intended to provide a layer of assurance between deployed AI systems and the functions responsible for organisational risk. The company combines structured testing before deployment with continuing monitoring and evidence that can be used when reviewing whether an application satisfies defined requirements.
Its technology is aimed particularly at financial services, insurance, healthcare and other organisations where automated activity may influence decisions involving sensitive information or regulated processes.
Testing continues after deployment
Before an AI agent enters production, Disseqt provides structured assessments intended to examine how the system responds to challenging instructions and whether its behaviour remains within specified operating limits. Such testing can identify weaknesses while developers still have opportunities to change the design.
However, a successful pre-deployment assessment cannot establish that the system will behave identically under every future condition. Business data changes, software is updated and users may provide requests that were not anticipated during development.
Disseqt therefore combines initial testing with runtime monitoring, intended to identify changes in behaviour and provide records that organisations can review. Its published product information also describes policy enforcement and continuing checks designed to support the operational management of AI systems.
The approach reflects a distinction between software functioning as intended during a controlled test and continuing to meet requirements once it is handling real business activity. An agent might perform accurately with a limited dataset but encounter different problems when processing records containing incomplete information or unusual exceptions.
Monitoring can help teams investigate those conditions, although its effectiveness depends on the information available from the agent and the connected systems. A monitoring platform cannot reliably identify every inappropriate action if significant parts of the workflow remain outside its visibility.
Disseqt’s proposed controls also need to operate alongside existing identity management, cybersecurity and application governance systems. Organisations deploying AI do not remove their established security responsibilities simply because an additional assurance platform records or evaluates agent behaviour.
Investor cites early operational results
Sure Valley Ventures has identified a customer example involving an AI agent used for fan engagement by a major international sports league. The investor says Disseqt improved the rate of responses satisfying defined safety criteria from 80% to 99%.
The example is associated with a sports league whose fan engagement operation has an audience of approximately 80 million people. That figure should not be interpreted as a verified count of individuals who interacted with the specific agent during the assessment.
Neither the underlying testing sample nor the full methodology for determining a safe response has been published in the material reviewed. The improvement therefore provides a company and investor-reported example rather than an independently validated benchmark for performance across other applications.
The investor also identifies work supporting a large European bank’s credit card programme. The published account does not name the bank or establish which functions have entered production, limiting the conclusions that can be drawn about deployment scale.
These examples indicate the types of organisations Disseqt is targeting, but its broader commercial position remains difficult to assess. The company has not disclosed a complete customer count, recurring revenue figure or independently audited results across its installed base.
Its investor-reported sales growth is similarly difficult to evaluate without knowing the starting level. A substantial percentage increase can reflect genuine commercial progress while providing limited information about the absolute size or profitability of a young software business.
Regulatory evidence becomes a product requirement
Disseqt’s proposition also responds to the need for organisations to document the controls applied to AI systems. In regulated sectors, businesses may need to explain how an automated application was tested, which risks were identified and what procedures are used to monitor its continuing operation.
The company refers to frameworks including ISO/IEC 42001 and the EU AI Act as part of its governance approach. Those frameworks can help organisations structure responsibilities and evidence, although the applicable legal duties depend on the nature of the AI system, the organisation’s role and how the technology is used.
Installing assurance software does not independently establish that a business complies with those obligations. Organisations remain responsible for determining which requirements apply and whether their controls are adequate for the risks presented by the system.
Disseqt is also working with systems integrators that deliver enterprise technology projects. Such relationships could allow assurance capabilities to be incorporated into AI deployments alongside application development and existing risk management procedures.
That delivery model may reduce the need for customers to introduce another isolated monitoring process, but its success will depend on whether the platform can work across different AI frameworks and technology environments without creating excessive operational complexity.
The company competes within a developing market that includes AI evaluation providers, cybersecurity vendors and established governance software businesses. Customers will need to assess which functions are already available within their technology estates and which risks require additional specialist tools.
Expansion will test the assurance model
Disseqt’s three founders bring experience from enterprise software, cybersecurity and commercial technology operations, while the company’s presence in Ireland, Britain and India gives it access to customers and development resources across several markets.
Individual investors in the latest round include current and former senior figures associated with major technology companies. Their personal investments should not be confused with corporate backing from those employers, and the company has not presented such participation as a commercial partnership with each organisation.
The new financing will support product development and customer expansion as more businesses move AI agents beyond limited trials. Commercial demand is likely to depend on whether assurance systems can provide evidence that security, compliance and operational teams find useful when deciding which automated processes should be permitted.
For Disseqt, the immediate challenge is demonstrating that its testing and monitoring work consistently across real enterprise deployments. The reported customer examples provide an early indication of its approach, while the investment will finance efforts to establish a broader and more independently assessable record of performance.












